Liver International. 2021;41:133–149. wileyonlinelibrary.com/journal/liv
|
133 Received: 14 February 2020|
Revised: 11 August 2020|
Accepted: 3 September 2020DOI: 10.1111/liv.14667 O R I G I N A L A R T I C L E
PDCD1 and IFNL4 genetic variants and risk of developing
hepatitis C virus-related diseases
Valli De Re
1| Maria Lina Tornesello
2| Mariangela De Zorzi
1| Laura Caggiari
1| Francesca Pezzuto
2| Patrizia Leone
3| Vito Racanelli
3|
Gianfranco Lauletta
3| Stefania Zanussi
1| Ombretta Repetto
1| Laura Gragnani
4| Francesca Maria Rossi
5| Riccardo Dolcetti
6| Anna Linda Zignego
4|
Franco M. Buonaguro
2| Agostino Steffan
1Valli De Re, Maria Lina Tornesello, these authors equally contributed to the study.
Abbreviations: BD, blood donors; B-NHL, B-cell non-Hodgkin lymphoma; CI, confidence interval; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MAF, minor allele frequency;
MC, mixed cryoglobulinaemia; OR, odds ratio; PD-1, programmed cell death protein 1; SNP, single nucleotide polymorphisms. 1Immunopathology and Cancer Biomarkers/
Bioproteomic facility, Department of Translational Research, Centro di Riferimento Oncologico (CRO) IRCCS, Cancer Institute, Aviano, Italy
2Molecular biology, viral oncology Istituto Nazionale Tumori IRCCS "Fondazione G. Pascale", Napoli, Italy
3Biomedical Sciences and Human Oncology, University of Bari Medical School, Bari, Italy 4Center for Systemic Manifestations of Hepatitis Viruses (MaSVE), Internal Medicine and Liver Unit, Department of Experimental and Clinical Medicine, Careggi University Hospital, Florence, Italy, Florence, Italy 5Clinical and Experimental Onco-Hematology Unit, Centro di Riferimento Oncologico (CRO) IRCCS, Aviano (PN), Italy 6The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Australia
Correspondence
Valli De Re, Immunopatologia e Biomarcatori Oncologici/Bio-proteomics Facility, Department of Research and Advanced Tumor Diagnostics, Centro di Riferimento Oncologico di Aviano (CRO), Via F. Gallini 2, I-33081 Aviano (PN), Italy.
Email: vdere@cro.it Funding information
FIRE/AISF ONLUS, Grant/Award Number: FIRE/AISF ONLUS (Fondazione Italiana per la Ricerc; 5X1000, Grant/Award Number: 5X1000_2010_MdS
Abstract
Background: Genetic variants of IFNL4 and PDCD1 genes have been shown to influ-ence the spontaneous clearance of hepatitis C virus (HCV) infection. We investigated the IFNL4 rs12979860 and the PDCD1 polymorphisms in 734 HCV-positive patients, including 461 cases with liver disease of varying severity and 273 patients with lym-phoproliferative disorders to determine the association of these genes with patient's outcome.
Methods: Expression levels of PDCD1 mRNA encoded by haplotypes were investi-gated by quantitative PCR in hepatocellular carcinoma (HCC) tissue and peripheral blood mononuclear cells. Flow cytometry was used to detect PD-1 and its ligand PD-L1.
Results: The frequency of IFNL4 rs12979860 C/T or T/T genotypes was signifi-cantly higher in patients with HCV-related diseases than blood donors (P < .0001). Patients expressing the IFNλ4 variant with one amino acid change that reduces IFNλ4 secretion was found increased in frequency in HCV-related diseases com-pared to HCC PDCD1 mRNA levels in HCC tissue were significantly higher in cases carrying the PD-1.3 A or the PD-1.7 G allele (P = .0025 and P = .0167). Linkage disequilibrium (LD) between PD-1.3 and IFNL4 was found in patients with mixed cryoglobulinaemia (MC) only (LD = 0 in HCC; LD = 72 in MC). PBMCs of MC pa-tients expressed low levels of PD-L1 in CD19+IgM+B cells and of PD-1 in CD4+T cells suggesting the involvement of regulatory B cell-T cell interaction to the patho-genesis of MC.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
1 | INTRODUCTION
Individuals with a hepatitis C virus (HCV) infection develop chronic hepatitis C (CHC) or remain asymptomatic. Recently, the incidence of hepatocellular carcinoma (HCC) has increased in association with
the increase in HCV infection.1 Chronic viral infection is
character-ized by dysfunctional B cells, which render some patients suscep-tible to the development of B lymphocyte proliferative diseases including mixed cryoglobulinaemia (MC) and B-cell non-Hodgkin
lymphoma (B-NHL).2,3 Interferon-lambda 4 (IFNL4) and programmed
cell death protein-1 (PD-1) expression levels have emerged as
im-portant modulators of altered host response to HCV infection 4–7
and as cause of a variety of autoimmune diseases.8 Immunotherapies
that inhibit the interaction between PD-1 and its ligands PD-L1 and PD-L2 have been shown to confer a substantial survival benefit in
patients with HCC9,10 as well as in those with haematological11 or
infectious diseases.12 The IFNL4 gene encodes the IFNλ4 protein
that increases the expression of IFN-stimulated genes (ISGs) in the liver. The receptors for IFNλ4 are expressed on selective cell types
like hepatocytes and some immune cells.13 IFNλ4 is expressed only
in individuals carrying the IFNL4 rs368234815 ∆G allele, which creates an open reading frame. In contrast, individuals with the rs368234815 T/T allele do not express the protein because of a premature stop codon. Other IFNL4 polymorphisms include a com-mon variant (rs117648444) resulting in a P70S amino acid change and two others (L79F and K154E), which lead to lower IFNλ4
se-cretion and antiviral activity.14,15 High IFNλ4 expression is strongly
associated with spontaneous HCV clearance and a lower response
to IFN-based treatment.16,17 The IFNL4 rs12979860[T] allele
ant may be used as an alternative to IFNL4 rs368234815 ΔG vari-ant to test for IFNL4 expression since their almost complete linkage
is well established.18 IFNλ4 was recently found to have selective
pressure on the HCV genome.19 Genome-wide association
stud-ies of patients with hepatitis infection identified polymorphisms in
IFNL4 and PDCD1 genes conferring a higher risk of chronic
infec-tion, HCV-related diseases and treatment response.8,20,21 Data from
patients with HCV-related lymphoproliferative diseases are limited, and there is a clear need to verify the role of IFNλ4 in the Caucasian population. The rs11568821 (PD-1.3) was associated with the de-velopment of autoimmune diseases in Europeans and Mexicans
but not in African Americans.22 In addition, the incidence of HCC
is higher among Asian Americans than White Americans23, while a
high incidence of HCV-related MC was present in the Mediterranean
basin,24 suggesting that genetic polymorphisms and environmental
risk factors may influence the risk of these diseases. Investigation of
IFNL4 and PDCD1 immunogenetic profiles in patients with different
HCV-related diseases could help to elucidate these pathologies. We therefore studied IFNL4 and PDCD1 genetic variants distribution and their relative expression in Italian patients diagnosed with different HCV-related diseases. Results could underlie the complex immune interactions leading to different clinical outcomes in chronic HCV-positive patients.
2 | PATIENTS AND METHODS
This study investigated data and biological samples obtained from 734 HCV-positive patients with different liver or lymphoprolifera-tive diseases and 94 non-HCV-infected blood donors. Liver diseases taken into consideration were CHC (n = 148), cirrhosis (n = 113) and HCC (n = 200), while the studied lymphoproliferative diseases were MC (n = 138) and NHL (n = 135). Patients were in treatment at one of the three participating Italian hospitals where they had been diag-nosed and had undergone HCV molecular analysis. For each patient,
Handling Editor: Benjamin Maasoumy Conclusion: Collectively, our data indicate an important contribution of IFNλ4
ex-pression to the development of HCV-related HCC and an epistatic contribution of IFNL4 and PDCD1 in MC.
Lay summary: Studies of IFNL4 and PDCD1 genes are helpful to better understand the role of host genetic factors and immune antigens influencing the outcome of HCV-related diseases. Our data support an association between the expression of IFNλ4, which prevents the expression of IFNλ3, with all the different HCV-related diseases studied, and besides, evidence that a higher IFNλ4 expression is associated with hepatocellular at a younger age. The expression pattern of low PD-L1 on B cells and high PD-1 on CD4+T-cells in patients with HCV-positive cryoglobulinaemia sug-gests a critical role of the PD-1/PD-L1 signaling in modulating B cell-T cell interaction in this lymphoproliferative disease.
K E Y W O R D S
we obtained from clinical records the following data: age at diagno-sis, sex, HCV positivity, HCV viral load and genotype, and specific diagnosis. Data on blood donors’ age and sex were collected.
In agreement with the Centre for Disease Controls and Prevention (CDC) guidelines, we defined as CHC a patient with long-lasting HCV infection who had persistent viraemia for more than 6 months after initial exposure diagnosed by blood test. Such test is based on the detection of serum anti-HCV antibodies per-formed by an enzyme immunoassay (III-generation EIA) against antigens from the HCV-core and from the HCV-nonstructural
regions and is confirmed by the qualitative analysis that detects the presence of HCV RNA (Cobas Amplicor HCV assay). All pa-tients we have included in the CHC group were positive to both these tests. The viral load was determined in 192 patients (Table 1) to monitor their response to treatment. Data reported in Table 1 were referred to tests performed before treatment. The viral load (branched DNA, Chiron, Emeryville, USA) was measured only in a subgroup of patients with HCV infection because in most cases a positive or a negative result was sufficient for clinical deter-minations. Sustained virological response (SVR) is defined as a TA B L E 1 Baseline characteristics of HCV patients and blood donors
Characteristics Age (years) BD CHC Cirrhosis HCC MC NHL 42 ± 10 57 ± 14 64 ± 11 67 ± 9 66 ± 10 68 ± 15 Sex male BD CHC Cirrhosis HCC MC NHL 90.1% 48.3% 65.4% 71.2% 29.3% 47.7% HCV positive viraemiaa BD CHCb Cirrhosis HCC MC NHL 0% 100% 100% 100% 100% 100% HCV viral loadb CHC Cirrhosis HCC MC NHL 2.5 ± 0.3 2.2 ± 0.4 1.9 ± 0.2 3.1 ± 0.2 3.0 ± 0.4 HCV genotypec CHC Cirrhosis HCC MC NHL genotype 1 60% 63% 58% 63% 60% genotype 2 26% 21% 28% 26% 32% genotype 3 13% 10% 9% 7% 4% genotype 4 2% 6% 4% 4% 4% CHC with
moderate fibrosis Mild-moderate fibrosis
60 40.5%
Severe fibrosis 59 39.9%
Data not available 29 19.6%
MC with
Mild-moderate fibrosis 27 19.6%
Severe fibrosis 24 17.4%
Data not available 87 75.0%
Note: Continuous variables reported as mean ± standard deviation. Categorical variables reported as %.
BD, blood donors, CHC, patients with a chronic HCV infection for more than 6 months; HCC, patients with hepatocellular carcinoma; MC, patients with mixed cryoglobulinaemia; NHL, patients with a non-Hodgkin lymphoma.
aHCV infection for more than 6 months confirmed at both serum anti-HCV antibodies and for the presence of HCV RNA. bData available for 192 cases;
negative response to serum HCV RNA viral load performed at least 12 weeks after the end of HCV treatment.
In addition, fibrosis stage assessment by liver biopsy (Metavir score) or fibroscan measure in patients with CHC and MC were retrieved when available (n = 119 CHC; n = 51 MC, Table 1). A good correlation was established between fibroscan and Metavir score as follow: F0-F1, 7.1 kPa; F2, 7.1-9.4 kPa; F3, 9.5-12.5 kPa; and F4, 12.5 kPa. We categorized patients with F0-F2 stage as a 'mild-moderate fibrosis' and those with a F3-F4 stage as 'advanced fibrosis'.
HCV genotypes have been determined using the Versant HCV Genotype 2.0 assay (Siemens Healthcare Diagnostics). A diagno-sis of HCC was based on the criteria of the European Association
for the Study of the Liver25 using non-invasive criteria, namely two
imaging techniques both demonstrating a focal lesion >2 cm in di-ameter with features of arterial hypervascularization. The diagnosis of cyroglobulinaemia was based on the detection of cryoglobulins, performed according to guidelines of the Associazione Italiana per la Lotta alle Crioglobulinemie. The diagnosis of NHL in the course of HCV infection has been histopathologically confirmed based on the
WHO classification.26
The study protocol was approved by the Comitato Etico Indipendente of the Azienda Ospedaliero-Universitaria Consorziale Policlinico di Bari, the Scientific Board and Ethics Committee of Fondazione G. Pascale Istituto Nazionale Tumori, the institutional review board code SPE 14.084 AOUC, Comitato Etico Area Vasta Centro AOU Careggi, Firenze and Comitato Etico Bio-banca CRO. The study was done in accordance with the Declaration of Helsinki. All re-search subjects provided written informed consent for the collection, storage and analysis of their samples and data. The analyses were car-ried out at the Centro di Riferimento Oncologico (Aviano, Italy) and Fondazione G. Pascale (Naples, Italy) in the period 2017-2019.
2.1 | Biological samples and nucleic acid extraction
At first diagnosis, patients had contributed a sample of venous blood for research purposes. Specifically, 461 patients were diagnosed with increasing severity of liver disease, from CHC towards cirrhosis and HCC, and 273 patients with MC or NHL (Table 1). Whole blood (2 mL) was obtained from each patient and cryopreserved at −20°C until use. Total genomic DNA was extracted from whole blood using DNeasy Kits (Qiagen). From most HCC patients, peripheral blood mononuclear cells (PBMCs) had been isolated (from 10 mL blood) by Ficoll-Hypaque density gradient centrifugation. HCC patients had also contributed part of a biopsy sample of liver tissue that was not needed for pathological examination; this part had been stored in RNAlater stabilizing solution at −80°C until use. Total DNA wasextracted from HCC biopsies according to published procedures.27
Total RNA was extracted from approximately 20 mg of frozen HCC
tissue and approximately 1 × 106 PBMCs using Trizol Reagent
fol-lowing manufacturer's instructions. Quality and quantity of isolated nucleic acids were assessed with the spectrophotometer Nanodrop
2000C (Thermo Fisher Scientific). Samples with ratio of absorbance at 260 nm and 280 nm equal to or above 1.8 were further analysed.
2.2 | Selection of known IFNL4 and PDCD1
polymorphisms and IFNL4 mutational analysis
For IFNL4, we chose to genotype rs12979860 because of its known linkage disequilibrium with rs368234815 which is required for IFNλ4
expression4 and because we had validated a method for rs12979860
genotyping in a previous study.17
For PDCD1, we selected single nucleotide polymorphisms (SNPs) known to be associated with susceptibility to cancer or autoimmune
diseases28–45 or to influence the spontaneous resolution of HCV
in-fection or the response to antiviral treatment46–49 (Table S1). From
this preliminary list of six SNPs, we selected four SNPs (ie PD-1.3, PD-1.5, PD-1.6 and PD-1.7) for subsequent analysis because they had a minor allele frequency (MAF) ≥0.05 in the Italian population according to the Ensemble website database (http://www.ensem bl.org/Homo_sapie ns/; Table S1).
To identify other genetic variations potentially affecting IFNλ4 activity, we sequenced the IFNL4 gene (intron 1 to exon 5) in ge-nomic DNA from PBMCs of 36 cases homozygous at rs12979860
(16 C/C and 20 T/T). Sequencing was done as previously described4
but with a modified sequencing primer (IFNL4 int IV1, Table S2). The amino acid changes G58R and P70S in the IFNλ4 protein flank residue N61, whose glycosylation is required for secretion of active
IFNλ4 protein.50 Consequently, these variants lower the antiviral
ac-tivity of the reference IFNλ4.51
2.3 | Genotyping and sequencing of
IFNL4 and PDCD1
Genotyping of IFNL4 rs12979860 and rs117648444 (C>T, P70S) was performed using custom TaqMan SNP genotyping assays (Applied Biosystems) on a 7900HT Fast Real-Time PCR system (Applied Biosystems).
To sequence IFNL4 (from intron 1 to exon 5), we first amplified the gene from genomic DNA by PCR in a reaction volume of 25 μl (200 ng dNTPs and 0.5 U GoTaq DNA Polymerase, Promega) using primers as reported in Table S2 and cycling conditions as described
in reference 4. Sequencing was performed as previously reported.52
To genotype PDCD1 SNPs, genomic DNA (30-300 ng) was am-plified in a 50 μL reaction mixture containing 10 pmol of each primer (Table S2), 1.25 U Hot Master Taq DNA Polymerase (5 Prime) and 25 μL PreMixJ (MasterAmp PCR, Epicentre, Madison, USA) on a Sure Cycler 8800 thermal cycler (Agilent Technologies). Amplification started with an initial denaturation at 94°C for 3 min, followed by 30 amplification cycles of denaturation at 94°C for 30 s, annealing at 65°C for 30 s, elongation at 72°C for 1 min, and a 10 min final elon-gation at 72°C. PCR products were subjected to Sanger automated sequencing analysis.
2.4 | LD analysis
Linkage disequilibrium (LD) was estimated among 1, that is, PD-1.3, PD-1.5, PD-1.7 and IFNL4 rs12979860 SNP. Distribution of the
Lewontin's coefficient D′ and correlation coefficient r2 was
calcu-lated as the measures of LD using the SHEsis online tool (http://analy sis.bio-x.cn). LD within each genomic region was explored and the extent of statistical significance of each pairwise association was represented by a scale of colour intensity.
2.5 | Analysis of PDCD1 mRNA expression
The expression levels of haplotypes within PDCD1 gene were evaluated by real-time quantitative PCR using total RNA from PBMCs of selected cases based on their PDCD1 haplotype (n = 12) and from selected HCC tissues (n = 12). Briefly, total RNA (500 g) was reverse transcribed in a 20 μL volume with Superscript II RNase H and an oligo d(T) primer (Life Technologies). The obtained cDNA (2 µL) was then subjected to qPCR using forward (5’-GAGGGACAATAGGAGCCAGG-3’) and reverse (5’-TCTTCTCTCGCCACTGGAAA-3’) primers targeting PDCD1 exons 4 and 5, respectively, to exclude the amplification of cross-contaminating DNA. Each reaction contained 12.5 μL of 1x iQ SYBR Green supermix (Bio-Rad Laboratories), 10 pmol of each primer, 2 μL of cDNA and nuclease-free water in a final volume of 25 μL. Amplifications were performed in triplicate using the CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories). Gene
ex-pression was analysed using the comparative Ct (2-ΔCt) method of
quantification and SRSF4 as reference sample. The ΔCt values for each transcript were calculated by subtracting the respective Ct
(cycle threshold) value from the corresponding SRSF4 Ct (ΔCt = Ctx
– CtSRSF4). Ct values were corrected for the efficiency of primer
pairs.53
2.6 | PD-1 and PD-L1 flow cytometry analysis
Cryopreserved peripheral blood mononuclear cells (PBMCs) from two negative patients with HCC, two patients with HCV-related HCC and two patients with HCV-HCV-related MC were used forPD-1 and PD-L1 flow cytometry analysis. Before the analysis, the
dimethyl sulfoxide (DMSO) cryoprotectant was removed by wash-ing and centrifugation. PBMCs were re-suspended in phosphate-buffered saline (PBS). Sample were stained with combination of vital dye 7-aminoactinomycin (7-AAD), CD4 PE (cl. SK3), PD-1 FITC (cl. MIH4), or CD19 APC (cl.HIB19), PD-L1 PE (cl.MIH1), all from Becton Dickinson, San Jose, CA) and IgM FITC (Dako). All samples were ac-quired on a FACSCantoII flow cytometer and DiVa software (Becton Dickinson). At least 1,000 T or B cells were acquired per tube. Media of percent positive cells from vital lymphocytes were calculated be-tween HCC HCV-negative, HCC HCV-positive or MC-HCV-positive samples.
2.7 | Statistical analysis
Fisher's exact test and ANOVA analysis of variance were used to compare allele and genotype frequency of gene polymorphisms between patient groups with different pathologies and control subjects.
Multivariate logistic regression analysis was performed with diagnosis as the dependent variable and genotype as indepen-dent variable; age and sex were valuated as covariables. Odds ratios and 95% confidence intervals were calculated. Genotypes of each polymorphism were assessed according to dominant (0 wild-type homozygote; 1 heterozygote and variant homozygote), recessive (0 wild-type homozygote and heterozygote; 1 variant homozygote) and additive genetic models (ie overdominant and log-additive).
The chi-squared test for trends was used to assess associations with genotype for liver diseases of increasing severity (chronic HCV infection, cirrhosis and HCC).
Statistical analyses were performed using MedCalc software (version 17.2), SNPStats (https://www.snpst ats.net/start.htm) and SHEsis online tool (http://analy sis.bio-x.cn). A P < .05 was taken to indicate statistical significance. Bonferroni for dependent SNPs and Sidak corrections for independent SNPs were utilized to conduct multiple comparisons test.
3 | RESULTS
This study examined the immunogenetic profiles of 734 HCV-infected patients, 461 with a liver disease (148 CHC, 113 cirrhosis, 200 HCC) and 273 with a lymphoproliferative disease (138 MC, 135 NHL) and 98 non-HCV-infected blood donors, all living in Italy (Table 1). Among the patients with liver disease, there were 148 with chronic HCV infection, 113 patients with cirrhosis and 200 patients with HCC. Among those with a lymphoproliferative disease, there were 138 with MC and 135 with B-NHL. Male sex predominated among cirrhosis and HCC patients, while female sex predominated among MC patients. Blood donors (BD) were mostly male. Fibrosis data were available for 119 patients included in the CHC group and for 51 patients included in the MC group (Table 1). Among the CHC group, 60 patients (40.5%) were diagnosed with mild-moderate liver fibrosis (fibrosis F1-F2) and 59 (39.9%) with severe fibrosis (F3-F4); in the MC group, 27 patients (19.6%) had mild-moderate liver fibrosis and 24 (17.4%) a severe fibrosis (F3-F4). Response data to anti-HCV treatment were available for 319 patients: 101 with CHC, 82 with cirrhosis, 34 with HCC, 66 with MC and 36 with NHL.
3.1 | IFNL4 and PDCD1 genotype frequencies
Patients and controls were genotyped for five SNPs having possible immune-modulating effects, including rs12979860 in IFNL4 and fourT A B LE 2 Fr eq ue nc ie s o f P D -1 a nd I FN L4 g en ot yp es i n p at ie nt s w ith H C V-re la te d d is ea se s a nd i n b lo od d on or s PD -1 .3 ( % ) rs 11 56 88 21 # PD -1 .5 ( % ) rs 222 79 81 # PD -1 .6 ( % ) rs 102 04 525 # PD -1 .7 ( % ) rs 74 218 61 # IF N L4 ( % ) rs 12 97 98 60 # B D ( n = 9 8) G /G 8 2 ( 84 % ) G /A 1 6 ( 16 % ) C /C 3 6 ( 37 % ) C /T 4 4 ( 45 % ) T/ T 1 8 ( 18 % ) C /C 7 9 ( 81 % ) C /T 1 9 ( 19 % ) T/ T 0 A /A 4 2 ( 43 % ) A /G 4 7 ( 48 % ) G /G 9 ( 9% ) C /C 4 4 / 45 % ) C /T 4 8 ( 49 % ) T/ T 6 ( 6% ) C H C ( n = 14 8) G /G 1 14 ( 77 % ) G /A 3 4 ( 23 % ) C /C 3 9 ( 26 % ) C /T 8 5 ( 57 % ) T/ T 2 4 ( 16 % ) C /C 1 25 ( 84 % ) C /T 2 0 ( 14 % ) T/ T 3 ( 2% ) A /A 78 ( 53 % ) A /G 5 6 ( 38 % ) G /G 1 4 ( 9% ) C /C 3 9 ( 26 % ) C /T 8 7 ( 59 % ) T/ T 2 2 ( 15 % ) A ) C om pa re d t o B D a C H C ( n = 14 8) G /G 1 14 ( 77 % ) G /A 3 4 ( 23 % ) C /C 3 9 ( 26 % ) C /T 8 5 ( 57 % ) T/ T 2 4 ( 16 % ) C /C 1 25 ( 84 % ) C /T 2 0 ( 14 % ) T/ T 3 ( 2% ) A /A 78 ( 53 % ) A /G 5 6 (3 8% ) G /G 1 4 ( 9% ) C /C 3 9 ( 26 % ) C /T 8 7 ( 59 % ) T/ T 2 2 ( 15 % ) lo g-add iti ve 2. 04 ( 1. 33 -3 .1 3) P < .000 1 C irr ho si s ( n = 11 3) G /G 9 2 ( 81 % ) G /A 2 1 ( 19 % ) C /C 3 4 ( 30 % ) C /T 5 8 ( 51 % ) T/ T 2 1 ( 19 % ) C /C 9 3 ( 82 % ) C /T 1 9 ( 17 % ) T/ T 1 ( 1% ) A /A 5 3 ( 47 % ) A /G 4 4 ( 39 % ) G /G 1 6 ( 14 % ) C /C 2 6 ( 23 % ) C /T 6 6 ( 58 % ) T/ T 2 1( 19 % ) dom ina nt 2. 73 ( 1. 51 -4 .9 3) P < .0 00 1 H C C ( n = 2 00 ) G /G 1 51 ( 75 % ) G /A 4 5 ( 23 % ) A /A 4 ( 2% ) C /C 7 4 ( 37 % ) C /T 9 4 ( 47 % ) T/ T 3 2 ( 16 % ) C /C 1 59 ( 80 % ) C /T 3 9 ( 20 % ) T/ T 2 ( 1% ) A /A 7 5 ( 38 % ) A /G 1 14 ( 57 % ) G /G 1 1 ( 5% ) C /C 5 1 ( 26 % ) C /T 1 03 ( 52 % ) T/ T 4 6 ( 23 % ) lo g-add iti ve 2. 28 ( 1. 55 -3 .3 5) P < .0 00 1 M C ( n = 13 8) G /G 9 9 ( 72 % ) G /A 3 9 ( 28 % ) rec es si ve 2. 02 ( 1. 05 -3 .8 7) P = .03 0 C /C 5 3 ( 38 % ) C /T 6 2 ( 45 % ) T/ T 2 3 ( 17 % ) C /C 1 09 ( 79 % ) C /T 2 8 ( 20 % ) T/ T 1 ( 1% ) A /A 5 5 ( 40 % ) A /G 6 3 ( 46 % ) G /G 2 0 ( 14 % ) C /C 5 8 ( 42 % ) C /T 6 1 ( 44 % ) T/ T 1 9 ( 14 % ) rec es si ve 2. 45 ( 0. 94 -6 .3 8) P = .0 53 N H L ( n = 13 5) G /G 1 13 ( 84 % ) G /A 2 2 ( 16 % ) C /C 4 1 ( 30 % ) C /T 6 0 ( 45 % ) T/ T 3 4 ( 25 % ) C /C 1 06 ( 79 % ) C /T 2 7 ( 20 % ) T/ T 2 ( 2% ) A /A 6 9 ( 51 % ) A /G 5 4 ( 40 % ) G /G 1 2 ( 9% ) C /C 4 5 ( 33 % ) C /T 6 6 ( 49 % ) T/ T 2 4 ( 18 % ) rec es si ve 3. 28 ( 1. 29-8. 37 ) P = .0 06 9 To ta l l iv er d is ea se s (n = 4 61 ) G /G 3 57 ( 77 % ) G /A 1 00 ( 22 % ) A /A 4 ( 1% ) C /C 1 47 ( 32 % ) C /T 2 37 ( 51 % ) T/ T 7 7 ( 17 % ) C /C 3 77 ( 82 % ) C /T 78 ( 17 % ) T/ T 6 ( 1% ) A /A 2 06 ( 45 % ) A /G 2 14 ( 46 % ) G /G 4 1 ( 9% ) C /C 1 16 ( 25 % ) C /T 2 56 ( 56 % ) T/ T 8 9 ( 19 % ) lo g-add iti ve 2. 23 ( 1. 56 -3 .1 9) P < .0 00 1 Ly m pho pr ol ifer at iv e di se as es ( n = 27 3) G /G 2 12 ( 78 % ) G /A 6 1 ( 22 % ) C /C 9 4 ( 34 % ) C /T 1 22 ( 45 % ) T/ T 5 7 ( 21 % ) C /C 2 15 ( 79 % ) C /T 5 5 ( 20 % ) T/ T 3 ( 1% ) A /A 1 24 ( 45 % ) A /G 1 17 ( 43 % ) G /G 3 2 ( 12 % ) C /C 1 03 ( 38 % ) C /T 1 27 ( 47 % ) T/ T 4 3 ( 16 % ) rec es si ve 2. 84 ( 1. 17 -6 .8 9) P = .0 11 A ll H C V-po si tiv e pa tie nt s ( n = 73 4) G /G 5 69 ( 78 % ) G /A 1 61 ( 22 % ) A /A 4 ( 1% ) C /C 2 41 ( 33 % ) C /T 3 59 ( 49 % ) T/ T 1 34 ( 18 % ) C /C 5 92 ( 81 % ) C /T 1 33 ( 18 % ) T/ T 9 ( 1% ) . A /A 3 30 ( 45 % ) A /G 3 31 ( 45 % ) G /G 7 3 ( 10 % ) C /C 2 19 ( 30 % ) C /T 3 83 ( 52 % ) T/ T 1 32 ( 18 % ) lo g-add iti ve 1. 86 ( 1. 33 -2 .6 0) P < .0 00 1 B) C om pa re d t o C H C a C irr ho si s ( n = 11 3) G /G 9 2 ( 81 % ) G /A 2 1 ( 19 % ) C /C 3 4 ( 30 % ) C /T 5 8 ( 51 % ) T/ T 2 1 ( 19 % ) C /C 9 3 ( 82 % ) C /T 1 4 ( 12 % ) T/ T 6 ( 5% ) A /A 5 3 ( 47 % ) A /G 4 4 ( 39 % ) G /G 1 6 ( 14 % ) C /C 2 6 ( 23 % ) C /T 6 6 ( 58 % ) T/ T 2 1( 19 % ) H C C ( n = 2 00 ) G /G 1 51 ( 75 % ) G /A 4 5 ( 22 % ) A /A 4 ( 2% ) Re ce ss iv e P = .0 35 C /C 7 4 ( 37 % ) C /T 9 4 ( 47 % ) T/ T 3 2 ( 16 % ) do mina nt 0. 61 (0. 38 -0. 97 ) P = .0 35 C /C 1 59 ( 80 % ) C /T 3 9 ( 19 % ) T/ T 2 ( 1% ) A /A 7 5 ( 38 % ) A /G 1 14 ( 57 % ) G /G 1 1 ( 5% ) dom ina nt 1. 86 ( 1. 21 -2 .8 6) P < .0 05 C /C 5 1 ( 26 % ) C /T 1 03 ( 52 % ) T/ T 4 6 ( 23 % ) (Co nti nue s)
SNPs in PDCD1 genes (Table S1). The PDCD1 SNPs, termed PD-1.3 (rs11568821), PD-1.5 (rs2227981), PD-1.6 (rs10204525) and PD-1.7 (rs7421861), had a MAF ≥0.05 in the reported Toscani population in Italy as well as in BD (Table S1). Genotype frequencies of the five SNPs are given in Table 2, in Table S3 according to fibrosis and in Table S4 according to SVR.
Multivariate logistic regression was done to identify the associ-ations between genotype and clinical group according to different genetic models (Table 2). The IFNL4 rs12979860[T] allele associ-ated with HCV positivity (n = 734) compared to BD (n = 98), in a log-additive model with an odds ratio (OR) of 1.86 (95% CI, 1.33-2.60, P < .0001), that remains significant after both Bonferroni's and Sidak's corrections. The association was more evident in patients with liver diseases, who had a lower frequency of the protective gen-otype C/C than patients with a lymphoproliferative disease (25% vs 38%; Table 2). T/T genotype showed a positive trend with the in-crease in liver disease severity (P = .018. Table S5) and apparently not affected by a sustained response to anti-HCV treatment (Table S6).
The four PDCD1 SNPs showed no association when geno-type frequencies were compared between all HCV-positive pa-tients and BD (Table 2). The analysis of individual clinical groups showed a higher frequency of PD-1.3 G/A genotype only among MC patients compared to BD (28% vs 16%); in a recessive model, OR = 2.02 (95% CI, 1.05-3.87; P = .030). These results suggest that PDCD1 polymorphisms have no effect on the clearance of HCV infection. Further examination revealed associations be-tween PDCD1 SNP genotypes and type of liver disease. In par-ticular, patients with HCC had a higher frequency of PD-1.5 C/C alleles (37% vs 26%, OR = 0.61, P = .035) and a lower frequency of PD-1.7 A/A (53% vs 38%, OR = 1.86, P < .005) compared to those with CHC. After Bonferroni's and Sidak’ s correction only PD-1.7 A/A association remained significant. The PD-1.7 A/A al-lele distribution in liver diseases has shown to follow a negative trend with disease severity meaning a lower frequency in HCC (38%) followed by cirrhosis (47%), CHC with severe fibrosis (51%) and CHC with mild-moderate fibrosis (52%); chi-squared test for trend for PD-1.7 = 5.596, P = .018 (Table S5). SVR did not signifi-cantly affect the genotype association (Table S6). Patients with a lymphoproliferative disease had a higher frequency of the PD-1.5 C/C allele (34%) than did patients with CHC (26%); in an overdom-inant model, OR = 0.60, P = .012, P-value remained statistically significant after both Bonferroni and Sidak's corrections.
3.2 | IFNL4 mutational analysis
The IFNL4 gene was sequenced from intron 1 to exon 5 in 36 cases homozygous at IFNL4 rs12979860 with either a C/C (n = 16) or T/T (n = 20) genotype. This analysis identified eight polymorphisms in 24 patients (Figure 1). Already known non-synonymous variants rs117648444 (P70S, five cases) and rs746231316 (G58R, one case) were identified in exon 2. The amino acids changed by these SNPs flank
residue N61, which glycosylation is required for functional IFNλ4.50
PD -1 .3 ( % ) rs 11 56 88 21 # PD -1 .5 ( % ) rs 222 79 81 # PD -1 .6 ( % ) rs 102 04 525 # PD -1 .7 ( % ) rs 74 218 61 # IF N L4 ( % ) rs 12 97 98 60 # M C ( n = 13 8) G /G 9 9 ( 72 % ) G /A 3 9 ( 28 % ) C /C 5 3 ( 38 % ) C /T 6 2 ( 45 % ) T/ T 2 3 ( 17 % ) do mina nt 0. 57 (0. 35 -0. 95 ) P = .029 C /C 1 09 ( 79 % ) C /T 2 8 ( 20 % ) T/ T 1 ( 1% ) A /A 5 5 ( 40 % ) A /G 6 3 ( 46 % ) G /G 2 0 ( 14 % ) lo g-add iti ve 1. 48 ( 1. 04 -2 .0 9) P = .0 26 C /C 5 8 ( 42 % ) C /T 6 1 ( 44 % ) T/ T 1 9 ( 14 % ) dom ina nt 0. 50 (0. 30 -0. 82 ) P = .0 06 N H L ( n = 13 5) G /G 1 13 ( 84 % ) G /A 2 2 ( 16 % ) C /C 4 1 ( 30 % ) C /T 6 0 ( 45 % ) T/ T 3 4 ( 25 % ) ov er do mina nt 0. 59 (0. 37 -0. 95 ) P = .029 C /C 1 06 ( 79 % ) C /T 2 7 ( 20 % ) T/ T 2 ( 2% ) A /A 6 9 ( 51 % ) A /G 5 4 ( 40 % ) G /G 1 2 ( 9% ) C /C 4 5 ( 33 % ) C /T 6 6 ( 49 % ) T/ T 2 4 ( 18 % ) Li ve r d is ea se s ex clu di ng C H C (n = 3 13 ) G /G 2 43 ( 78 % ) G /A 6 6 ( 22 % ) A /A 4 ( 1% ) C /C 1 08 ( 34 % ) C /T 1 52 ( 49 % ) T/ T 5 3 ( 17 % ) C /C 2 52 ( 81 % ) C /T 5 8 ( 18 % ) T/ T 3 ( 1% ) A /A 1 28 ( 41 % ) A /G 1 58 ( 50 % ) G /G 2 7 ( 9% ) do mina nt 1. 61 ( 1. 09 -2 .3 9) P = .02 C /C 7 7 ( 25 % ) C /T 1 69 ( 54 % ) T/ T 6 6 ( 21 % ) Ly m pho pr ol ifer at iv e di se as es e xc lu de d C H C ( n = 27 3) G /G 2 12 ( 78 % ) G /A 6 1 ( 21 % ) C /C 9 4 ( 34 % ) C /T 1 22 ( 45 % ) T/ T 5 7 ( 21 % ) ov er dom ina nt 0. 60 (0. 40 -0. 90 ) P = .0 12 C /C 2 15 ( 79 % ) C /T 5 5 ( 20 % ) T/ T 3 ( 1% ) A /A 1 24 ( 45 % ) A /G 1 17 ( 43 % ) G /G 3 2 ( 12 % ) C /C 1 03 ( 38 % ) C /T 1 27 ( 47 % ) T/ T 4 3 ( 16 % ) do mina nt 0. 59 (0. 38 -0. 92 ) P = .0 17 A bb re vi at io ns : B D , b lo od d on or s; C H C , c hr on ic H C V i nf ec tio n; H C C , h ep at oc el lu la r c ar ci no m a; M C , a ut oi m m un e l ym ph op ro lif er at iv e m ix ed c ry og lo bu lin ae m ia ; N H L, n on -H od gk in l ym ph om a. Th e m os t s ig ni fic an t m od el f or a ss oc ia tio n w ith d ia gn os is ( O R ( 95 % C I) i n c om pa ris on w ith B D i s r ep or te d o nl y w he n P v al ue i s s ig ni fic an t ( P < .0 5) . A B on fe rr on i's c or re ct io n f or m ul tip le S N Ps r es ul ts i n a P-va lu e t hr es ho ld o f 0 .0 12 5 a nd o f S id ak 's c or re ct io n o f 0 .0 12 7. S ig ni fic an t S N Ps a ft er c or re ct io ns a re i n b ol d t ex t. a T A B LE 2 (Co nti nue d)
Consequently, these variants reduce the protein antiviral activity.51 Our
results showed that these two variants associate with the rs12979860 T/T genotype (six of 20 cases, 30%) only. Finally, a new synonymous mutation was identified in exon 3 (one case), and the known synony-mous variant rs12971396 was found in exon 5 (16 cases). This latter SNP associated with the rs12979860 T/T genotype (15 of 20, 75%). Although based on a small number of patients, these data suggest an association between HCC and secretion of fully active IFNλ4 (patients homozygous for rs12979860[T] without a G58R or P70S variant in
IFNL4; Table 3). Indeed, the rs12979860[T] polymorphism necessary
for IFNλ4 production is associated with HCC (Table 2) and the fre-quency of the genotype necessary for fully active IFNλ4 expression is higher in HCC (no G58R or P70S variant in 15 of 28, 54%) than in all other HCV-positive patients (17 of 48, 35%) (Table 3). HCC patients with a fully active IFNL4 expression were diagnosed with tumour at younger age than those with reduced expression (Figure 2). These data suggest that IFNλ4 levels are a risk factor for HCC development.
3.3 | PDCD1 polymorphisms and mRNA expression
We next examined whether PDCD1 polymorphisms affected mRNA expression in tumour biopsies and PBMCs from HCC pa-tients (Figure 3A). We found higher PDCD1 mRNA expression in HCC tissues carrying the PD-1.3 A/G genotype than the G/A genotype (P = .0025, Mann-Whitney test; Figure 3B and C). Similarly, expression was higher in HCC tissues with the PD-1.7 G polymorphism (A/G and G/G genotypes) than with the A/A genotype (P = .0167, Mann-Whitney test; Figure 3B and C). Conversely, no significant difference in PDCD1 mRNA expres-sion was observed among PBMCs from HCC patients with dif-ferent PD-1 genotypes (Figure 3C). Considering the important role of the PD-1–PD-L1 axis in regulating T-cell function, these results suggest that a high expression of PD-1 in the HCC mi-croenvironment contributes to hepatocarcinogenesis by locally impairing antitumour immunity.F I G U R E 1 Genomic positions of IFNL4 genetic variants in samples homozygous at rs12979860. In addition to rs12979860, eight different SNPs were identified in 24 of 36 samples. Three SNPs (including one in exon 3) have not previously been reported. The table below the gene diagram indicates the genotype at each polymorphism for all 36 samples
3.4 | Epistatic interactions between IFNL4 and
PDCD1 variants
Haplotype data for the PDCD1 SNPs PD-1.3, PD-1.5 and PD-1.7 and for rs12979860 of IFNL4 were analysed using SHEsis software,
yielding eight common haplotypes with a frequency >5% (Table 4). In addition, a comparison between cirrhosis and HCC has been per-formed (Table S7). These results suggested that specific haplotypes may influence the susceptibility to or progression of different HCV-related diseases (Table 4). In addition, linkage disequilibrium analy-sis (LD) evidenced a linkage that did not occur by chance (LD >50) between PDCD1 SNPs and rs12979860 of IFNL4 in BD and in pa-tients with CHC, MC and B-NHL but not in papa-tients with cirrhosis or HCC. An association between LD-1.3 and IFNL4 was found in MC (LD = 72) but not in HCC (LD = 0) (Figure 4). In addition, real-time polymerase chain reaction (PCR) indicated higher PDCD1 expression in HCC tissue samples carrying a PD-1.3 G/A allele and a PD-1.7 G/A or G/G allele than in those with A/A genotypes, respectively, but not in PBMCs of the same patients (Figure 3).
Overall these results underline the importance of IFNL4 ex-pression (IFNL4 rs12979860[T] variant) and low hepatic exex-pression of PD-1, found in the GCAT haplotype, in predicting HCV-related HCC (OR = 2.101; 95% CI, 1.265-3.491, Table 4). Conversely, ex-pression of PDCD1 (ACGC haplotype) in addition to the more po-tent anti-HCV IFNL3 expression (IFNL4 rs12979860[C] carrier) was in large part associated with MC (OR = 4.237; 95% CI, 1.923-9.334). Moreover, when IFNλ4 was expressed (IFNL4 rs12979860[T] car-rier), its antiviral effect in MC patients was frequently reduced by the presence of a G58R or P70S variant in IFNλ4, by lowering the level of IFNλ4 secretion (present in 69% of patients carrying the
IFNL4 rs12979860[T] allele in our series). When IFNλ4 is produced
in MC, PDCD1 expression seems less important (GCAT haplotype; OR = 1.895; 95% CI, 1.097-3.275). Overall, these data suggest that the PD-1 SNPs, coupled with a low antiviral effect as a result of a reduced IFNL4 expression, have an important role in MC while ex-pression of a fully active IFNL4 (IFNL4 rs12979860[T] allele without G58R or P70S mutation) is more important in determining hepatic outcomes.
3.5 | PD-1 expression on T cells and PD-L1
expression on B lymphocytes of MC patients
A new B-cell subpopulation has recently been described, namely
CD19+ PD-L1+ regulatory B cell, which requires PD-L1 expression
to regulate CD4+ PD-1+ T follicular helper (Tfh) cell expansion and
differentiation and to suppress autoimmune diseases.54 Also in
pe-ripheral tissues, and particularly during chronic pathological immune
situations, CD4+ helper T cells belonging to follicular helper lineage
have an important role in stimulating B-cell responses, including plasma cell differentiation and production of high-affinity antibod-ies.55,56 Given that PD-L1hi B cells limit both memory B-cell
develop-ment and plasma cell differentiation (pathognomonic signs of MC)
by interacting with PD-1hi T cells,56,57 we analysed the expression of
PD-1hi in CD4+ T cells and of PD-L1hi in IgM+ CD19+ B lymphocytes
from patients with HCC without HCV infection (n = 2), with HCV-related HCC (n = 2) and with HCV-HCV-related MC (n = 2). We observed
>10% levels of PD-1hi expression on CD4+ T cells and ≥1% levels of
TA B L E 3 Frequency of IFNL4 genetic variants associated with IFNλ4 in patients and blood donors
IFNλ4 expression Fully Reduced patients with rs12979860-TT genotype tested for G58R or P70S IFNL4 variants G58R or P70S mutation Absent Present BD (n = 5) 3 (60%) 2 (40%)
Patients with liver
disease (n = 47) CHC (n = 11)Cirrhosis (n = 8) HCC (n = 28) 5 (45%) 2 (25%) 15 (54%) 6 (55%) 6 (75%) 13 (46%) Patients with lymphoproliferative disease (n = 29) All HCV-positive patients excluded HCC (n = 48) MC (n = 16) NHL (n = 13) 5 (31%) 5 (38%) 17 (35%) 11 (69%) 8 (62%) 31 (65%)
Abbreviations: BD, blood donors; CHC, chronic HCV infection; HCC, hepatocellular carcinoma; MC, autoimmune mixed cryoglobulinaemia; NHL, non-Hodgkin lymphoma.
F I G U R E 2 Boxplot describing the relationship of IFNL4 genotype related to a fully active IFNL4 secretion with a younger median age of HCC patients. HCC patients carrying a fully active (rs12979860-T/T combined without P70S mutation, rs117648444 C/C) were diagnosed with tumour at younger age compared to those with a reduced expression (rs12979860-T/T combined with P70S mutation, rs117648444 C/T or T/T), median age 61y and 72y, respectively, P = .02 ANOVA test. Boxes range from the 25th to the 75th percentile with a horizontal black line at the median and vertical lines extending to the 10th and 90th percentiles. These data suggest that IFNL4 levels may represent a risk factor for HCC development
PD-L1hi on IgM+ CD19+ B cells in HCC patients without HCV
infec-tion (Figure 5A), but lower than 10% levels of PD-1hi expression on
CD4+ T cells and absence of PD-L1 hi in patients with HCV-related
MC (Figure 5B).
4 | DISCUSSION
This study confirms the significant associations between the
IFNL4 rs12979860[T] variant and both persistence of HCV
in-fection and risk of HCC.17,51,58 The IFNL4 rs12979860[T] variant
is in high LD with the rs3682348[ΔG] polymorphism necessary for producing the antiviral cytokine IFNλ4 while preventing the
production of IFNλ3,51,59 which has a stronger antiviral activity
than IFNλ4. We identified in our series four SNPs in LD with the rs12979860[T] variant: two functional G58R and P70S variants
whose minor allele decreased IFNλ4 secretion,15 the intronic (IV3)
rs111531283[G] variant of unknown role but present in all cases,
and the synonymous rs12971396[C] variant in exon 5.15 The
fre-quencies of polymorphic alleles encoding a fully active IFNλ4 and a reduced secretion of IFNλ4 (ie G58R and P70S) were not signifi-cantly different between HCC patients and BD. Nonetheless, pa-tients with variants leading to a higher IFNL4 production showed a delay in HCC development (median age, 72 vs 60 years, P = .02). A possible explanatory hypothesis is that fully active IFNλ4-P70 causes a low level of viral replication, which in turn leads to an in-efficient adaptive immune response and consequently poor HCV
clearance60 and increased risk of developing HCC.
PD-1 inhibitors have been approved for patients with HCC,
and several pivotal trials are ongoing.61 We found that PD-1.7
(rs7421861) polymorphism after statistical Bonferroni's correc-tion maintains the significant associacorrec-tion with the risk of HCC compared with CHC. Our results suggest that PD-1 is unrelated to chronic infection because the polymorphic frequency of the same allele is similar in CHC and BD but might influence the clini-cal outcome of HCV infection. We examined the relation between
PDCD1 polymorphisms and mRNA expression in liver biopsies and
PBMCs of selected patients with HCC. Interestingly, HCC biopsies from patients carrying the 1.3 (rs11568821-A allele) and PD-1.7 (rs7421861-G allele) polymorphisms expressed higher levels of
PDCD1 mRNA; this association was not found in PBMCs from the
same patients. These results suggest that the HCC microenviron-ment promotes PDCD1 expression, whose level could be related to the PD-1.3 and PD-1.7 genotypes.
Since it is widely accepted that mutations have different effects in combination than individually, we analysed the relationship be-tween the PDCD1 haplotype and IFNL4 in HCC and compared it to that in patients with CHC. The GCAT haplotype, exhibiting lower
PDCD1 mRNA expression in tumour biopsies and a fully active IFNλ4
protein, associated with HCC in our series. Moreover, the ACGC haplotype, exhibiting higher PDCD1 mRNA and a lower IFNλ4 pro-tein expression, discriminates cirrhosis from HCC. Furthermore, LD analysis indicated that PD-1.3 and PD-1.7 are in linkage (LD = 72) in HCC and that IFNL4 is an independent gene (LD = 0). The ex-pression of PDCD1 in HCC cells is not well characterized. A recent study reported that 1 participates in the mTOR pathway in
PD-1-positive tumour cells.62 However, the specific biological role and
target implications of PD-1-positive cells in HCC biopsies require further investigation.
The association of IFNL4[TT] genotype in patients with HCV-related lymphoproliferative disorders was less evident than in patients with liver diseases. Furthermore, the association with HCV-related lymphoproliferative disorders was lower when consid-ering the less active form of IFNλ4 (IFNL4 rs12979860[T] coupled
to P70 mutation), which causes an increase in viral replication.60
We hypothesize that HCV replication intensifies the antigen-driven
immune stimulation that in turn sustains B-cell proliferation.63
The LD analysis showed an epistatic contribution between IFNL4 and PD-1.3 polymorphisms in MC patients (LD = 72). The study of haplotypes indicated that PD-1[ACG]-IFNL4[C] was positively asso-ciated with MC (OR = 4.237) and that the opposite haplotype [GTA] [T] (OR = 0.458) is negatively associated with MC. These results suggest that that PD-1.3 G allele may produce a PD-1 molecule that counterbalances the absence of IFNλ4 protein (rs12979860 = CC) in the risk of MC and vice versa.
Previous studies have demonstrated that high expression of PD-1 in hepatic lymphocytes, especially exhausted T cells and Tregs, is associated with a dysfunctional immune response in chronic HBV
infection and HCC,64–67 and that PD-1 had influence on the viral
profile.68 Studies on CHC 69,70 showed that overexpression of PD-1
on HCV-specific CD8+T cells was associated with a reduced effi-ciency of T cell–mediated cytolysis compared to non-HCV–specific T cells, and correlated with the maintenance of an exhausted
phe-notype of CD8+T cells.71 Notably, blockade of PD-1–PD-L1
interac-tions restored the activity of HCV-specific T cells,72 and controlled
HCV replication in a chimpanzee model of CHC.73
Lymphoid follicles with a germinal centre architecture were
commonly observed in the livers of patients with HCV infection.74
F I G U R E 3 PDCD1 mRNA expression based on PDCD1 haplotype. A, mRNA levels in HCV-positive HCC biopsy and PBMC from each patients (n = 12) with different PDCD1 polymorphisms. LC-297 to LC-433 are the identifier code of patients. Under the column it is reported
the haplotype of each single patient. The expression of PDCD1 gene was expressed in the y-axis as the 2−ΔCt value. The ΔCt values for
each amplified transcript were calculated by subtracting the respective Ct value from the corresponding control gene SRSF4 Ct (ΔCt = Ctx
– CtSRSF4). B, Median level of PDCD1 mRNA in HCC tissue and PBMC of patients stratified based on PD-1.3 and PD-1.7 genotype,
Mann-Witney test. C, Comparison of different expression level of PDCD1 mRNA in HCV-positive HCC tissue and PBMC from patients based on their PDCD1 genotype (PD-1.3, PD-1.5, PD-1.6. PD-1.7). The patients having PD-1.3 allele A and PD-1.7 allele G exhibited a higher PD-1 mRNA expression levels than those with wild-type homozygote (PD-1.3 genotype G/G, P = .0025 and PD-1.7 genotype A/A, P = .0182, Mann-Witney test respectively)
T A B LE 4 Ep is ta tic i nt er ac tio n d ef in ed b y PD -1 a nd IF N L4 p ol ym or ph is m s a nd t he ir a ss oc ia tio ns w ith H C V-re la te d d is ea se s c om pa re d t o p at ie nt s w ith a c hr on ic H C V i nf ec tio n Ha plo ty pe Fr eq uen cy Fr eq uen cy PD IF N L4 C irr ho sis (fr eq ) C H C (fr eq ) Χ 2 P value O dd s r at io [ 95 % C I] H CC (fr eq ) C H C (fr eq ) Χ 2 P value O dd s r at io [ 95 % C I] 1.3 1. 5 1.7 A C G C 21 .0 0 (0.0 53 ) 8. 29 ( 0. 02 8) 2. 82 0. 09 1. 99 2 [ 0. 878 -4 .5 19 ] A C G T 3. 11 ( 0. 01 4) 17 .5 0 ( 0. 05 9) 6. 25 0. 01 2 0. 237 [0.0 70 -0. 80 3] 20. 51 (0.0 51 ) 17 .5 0 ( 0. 05 9) 0.1 2 0. 73 0. 890 [0. 46 1-1. 71 7] G C A C 27 .6 0 ( 0. 12 2) 47 .6 0 ( 0. 16 1) 0. 90 0. 343 0. 78 2 [ 0. 47 0-1. 30 1] 56 .1 7 (0 .1 40 ) 47 .6 0 ( 0. 16 1) 0. 32 0. 57 0. 88 5 [ 0. 58 1-1. 34 8] G C A T 27 .3 2 (0 .1 21 ) 23 .2 0 ( 0. 078 ) 3. 59 0. 058 1. 75 3 [ 0. 97 6-3. 15 0] 58 .5 6 ( 0. 14 6) 23 .2 0 ( 0. 078 ) 8. 49 0.0 0 2. 10 1 [ 1. 26 5-3 .4 91 ] G C G C 37 .6 6 ( 0. 16 7) 33 .1 3 (0 .1 12 ) 4. 56 0. 03 3 1. 73 0 [1 .0 42 -2 .8 73 ] 43 .2 2 ( 0. 10 8) 33 .1 3 (0 .1 12 ) 0 0.9 9 0. 99 7 [0 .61 6-1. 61 3] G C G T 26 .6 3 ( 0. 11 8) 25 .0 8 (0.0 85 ) 2. 31 0.1 28 1. 56 2 [ 0. 87 6-2. 78 6] 36 .9 6 ( 0. 09 2) 25 .0 8 (0.0 85 ) 0. 23 0.6 3 1. 14 0 [ 0. 67 0-1. 94 1] G T A C 46 .5 3 ( 0. 20 6) 74 .5 8 ( 0. 25 2) 0. 67 0.4 03 0. 83 5 [ 0. 54 8-1. 27 4] 78 .78 ( 0. 19 7) 74 .5 8 ( 0. 25 2) 2. 29 0. 13 0. 75 6 [0. 52 7-1.0 87 ] G T A T 30 .6 9 ( 0. 13 6) 58 .4 2 ( 0. 19 7) 2. 35 0. 126 0. 68 8 [ 0. 42 5-1. 11 2] 61 .2 6 ( 0. 15 3) 58 .4 2 ( 0. 19 7) 1.7 9 0. 18 0. 76 3 [0 .5 13 -1 .13 5] PD IF N L4 M C ( fr eq ) C H C ( fr eq ) Χ 2 P value O dd s r at io [ 95 % C I] N H L (fr eq ) C H C ( fr eq ) Χ 2 P value O dd s r at io [ 95 % C I] 1.3 1. 5 1.7 A C G C 29 .4 9 ( 0. 10 7) 8. 29 ( 0. 02 8) 14 .8 2 <0 .0 01 4. 23 7 [ 1. 92 3-9. 33 4] A C G T 1. 71 ( 0. 00 6) 17 .5 0 ( 0. 05 9) 12 .12 <0 .0 01 0. 10 0 [0.0 21 -0. 48 8] 8. 23 ( 0. 03 0) 17 .5 0 ( 0. 05 9) 2. 67 0. 10 22 85 0. 50 0 [ 0. 21 4-1. 16 5] G C A C 31 .3 7 ( 0. 11 4) 47 .6 0 ( 0. 16 1) 2. 45 0. 11 8 0. 68 0 [ 0. 41 8-1. 10 5] 38 .7 1 ( 0. 14 3) 47 .6 0 ( 0. 16 1) 0. 33 0. 56 59 05 0. 87 3 [0 .5 50 -1 .3 87 ] G C A T 37 .6 4 ( 0. 13 6) 23 .2 0 ( 0. 078 ) 5. 38 0. 02 08 1. 89 5 [ 1. 09 7-3. 27 5] 27 .9 2 ( 0. 10 3) 23 .2 0 ( 0. 078 ) 1. 09 0. 29 612 2 1. 36 0 [0 .7 62 -2 .4 27 ] G C G C 41 .9 8 ( 0. 15 2) 33 .1 3 (0 .1 12 ) 2. 25 0. 13 4 1. 45 2 [ 0. 89 0-2. 37 0] 36 .3 3 ( 0. 13 5) 33 .1 3 (0 .1 12 ) 0. 69 0. 40 75 68 1. 23 7 [0 .7 47 -2 .0 50 ] G C G T 25 .8 1 ( 0. 09 4) 25 .0 8 (0.0 85 ) 0.1 8 0. 669 1. 13 4 [ 0. 63 7-2. 01 9] 18 .1 4 ( 0. 06 7) 25 .0 8 (0.0 85 ) 0. 61 0. 43 27 26 0. 778 [ 0. 41 4-1. 45 9] G T A C 68 .6 0 ( 0. 24 9) 74 .5 8 ( 0. 25 2) 0 0.9 90 1. 00 3 [ 0. 68 5-1. 46 7] 66 .7 5 ( 0. 24 7) 74 .5 8 ( 0. 25 2) 0.0 2 0. 90 08 48 0. 97 6 [0. 66 4-1. 43 4] G T A T 27 .6 0 ( 0. 10 0) 58 .4 2 ( 0. 19 7) 10 .17 0. 00 1 0. 45 8 [0. 28 1-0. 74 6] 58 .62 (0 .2 17 ) 58 .4 2 ( 0. 19 7) 0. 35 0. 554 954 1. 13 1 [0 .7 51 -1 .7 05 ] N ote : M ul ti-lo ci g en ot yp e f re qu en cy w ith a f re qu en cy < 0. 05 i n b ot h c on tr ol a nd c as es h as b ee n d ro pp ed . P v al ue s ig ni fic an t a t B on fe rr on i's c or re ct io n ( P-va lu e t hr es ho ld o f 0 .0 12 5) a re i n b ol d t ex t a nd hi gh lig ht ed i n r ed w he n p os iti ve a ss oc ia tio n w as f ou nd a nd i n g re y w he n a n eg at iv e a ss oc ia tio n w as f ou nd . O R ( 95 % C I), O dd s r at io w ith 9 5% c on fid en ce i nt er va l. C IR RH O SI S: G lo ba l c hi 2 i s 1 8. 33 00 76 w hi le df = 6 ; F is he r's P v al ue i s 0 .0 05 51 6. H C C : G lo ba l c hi 2 i s 1 3. 96 33 98 w hi le df = 7 ; F is he r's P v al ue i s 0 .0 52 06 1. M C : G lo ba l c hi 2 i s 4 3. 07 02 74 w hi le df = 7 ; F is he r's P v al ue i s 3 .4 8e -0 07 . N H L: G lo ba l c hi 2 i s 5 .2 51 82 3 w hi le df = 6 ; F is he r's P v al ue i s 0 .5 12 02 6. A bb re vi at io ns : B D , b lo od d on or s; C H C , c hr on ic h ep at iti s C v iru s i nf ec tio n; H C C , h ep at oc el lu la r c ar ci no m a; h ep at ic : c irr ho si s a nd H C C ; l ym ph : M C a nd N H L; M C , a ut oi m m un e l ym ph op ro lif er at iv e m ix ed cr yo gl ob ul ina emi a; N H L, n on -H odg ki n ly m ph om a.
Recently, a new B-cell subpopulation, the CD19+ PD-L1+
reg-ulatory B cell (Breg), has shown to require PD-L1 expression to
regulate CD4+ PD-1+ T follicular helper (Tfh) cell expansion and
differentiation and to mediate humoural immunity.54 These
find-ings disclosed novel mechanisms by which PD-1–PD-L1 signalling regulates antibody production and helps to understand the role of this pathway in physiological states and in the altered humoural immunity found in some autoimmune diseases. Given that
PD-L1hi B cells limit both memory B-cell development and plasma cell
differentiation by interacting with Tfh cells,56,57 we analysed the
expression of PD-1hi in CD4+ T cells and PD-L1 hi in IgM+ CD19+
B lymphocytes from patients with HCV-related MC. We observed
higher levels (>10%) of PD-1hi expression on CD4+T cells and
lower levels of PD-L1hi (≤1%) on IgM+ CD19+ B cells in patients
with HCV-related MC compared to patients with HCC without HCV-infection, suggesting an alteration in the control of B cell-T
cell interaction in our MC cases. These data are consistent with a
model in which the reduction in both CD4+PD-1hi + T cells and
regulatory B cells (PD-L1hi+), necessary to inhibit the signal of
ter-minal plasma cell differentiation and memory B-cell development,
is less effective in MC cases.54,75 The favourable actions of CD19+
PD-L1+ B cells in humoural B-cell homeostasis and in the control of
autoimmune diseases were further supported by the
demonstra-tion that these B cells are resistant to αCD20 B-cell depledemonstra-tion54
and that MC patients also benefit from αCD20 treatment.76
This is the first study evaluating a possible association between
PDCD1 polymorphisms and the risk of HCV-related
lymphoprolifer-ative disorders. These results should be considered descriptive, and larger studies are needed to confirm the model of IFNL4 and PDCD1 epistatic interactions in HCV-related MC and the independent con-tribution of the IFNL4[T] variant in the control of HCV infection and HCC development.
F I G U R E 4 Pairwise linkage
disequilibrium (LD) relationships between the PD-1.3, PD-1.5, PD-1.7 and IFNL4 rs12979860 polymorphism. A–F, Results from linkage analysis conducted for the polymorphism of blood donors (A), CHC (B), cirrhosis (C), HCC (D), MC (E) and NHL (F) respectively
Elimination of an infectious agent has been postulated to favour
the development of autoreactivity.77 Based on our results we
pro-posed that genes encoding immune proteins involved in the control of HCV infection, for example IFNλ4, need genes that encode other inhibitory proteins, for example PD-1, to inhibit the progression of the disease towards autoimmune MC. The previous study describ-ing that the 15-year cumulative probability of developdescrib-ing cirrho-sis and HCC was higher in MC(−) than in MC(+) patients (24.9% vs
14.2% and 20.3% vs 7.5% respectively),78 supports this model. In
keeping with these findings, we found different haplotype asso-ciations between HCC and MC patients, although further studies across different populations and functional assessments of relevant polymorphisms are required to confirm the associations of patho-genic relevance.
In conclusion, our study found that IFNL4 and PDCD1 polymor-phisms are both important in determining the risk of HCV-related MC, although we have yet to determine precisely how the proteins confer this risk. Our data also confirm and underline the important role of INFλ4 production as a risk factor for HCV persistence and HCC development. As a result of the importance of these genes in the immune response to hepatic infection, autoimmune disorders and malignancies, as well as their role in response to the emerging immune checkpoint treatment for HCC, our study adds new informa-tion to help understand the pathogenic role of host genetic variants in HCV-related diseases.
FUNDING INFORMATION
Mariangela De Zorzi, Laura Caggiari and Ombretta Repetto had fellow-ships funded by 5X1000_2010_MdS. Francesca Pezzuto is the recipient
of a research fellowship awarded by FIRE/AISF ONLUS (Fondazione Italiana per la Ricerca in Epatologia) http://www.fonda zione fegato.it/. ETHIC S STATEMENTS
This study was in accordance with the principles of Declaration of Helsinki and all subjects provided written informed consent. The study protocol was accepted by the Comitato Etico Indipendente of the Azienda Ospedaliero-Universitaria Consorziale Policlinico di Bari, the Scientific Board and Ethics Committee of Fondazione G. Pascale Istituto Nazionale Tumori, the Comitato Etico Area Vasta Centro AOU Careggi, Florence and Committee for the Bio-banking Facility of the Centro di Riferimento Oncologico di Aviano.
CONFLIC T OF INTEREST
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
ORCID
Valli De Re https://orcid.org/0000-0001-6100-9373
Maria Lina Tornesello https://orcid.org/0000-0002-3523-3264
Mariangela De Zorzi https://orcid.org/0000-0002-3794-270X
Laura Caggiari https://orcid.org/0000-0003-1651-6653
Francesca Pezzuto https://orcid.org/0000-0002-9585-6834
Patrizia Leone https://orcid.org/0000-0002-0904-1074
Vito Racanelli https://orcid.org/0000-0002-8639-1940
Gianfranco Lauletta https://orcid.org/0000-0003-0152-9398
Stefania Zanussi https://orcid.org/0000-0003-0608-9766
F I G U R E 5 PD-1 and PD-L1 flow cytometry analysis. The lymphocyte population was selected on a forward- versus side-scatterplot for each patient. PBMCs from HCV-negative HCC (n = 2), associated HCC (n = 2) and HCV-associated MC cases (n = 2) were analysed by flow cytometry after staining with CD4 and PD-1 antibodies, or anti-CD19, anti-IgM and anti-PD-L1 antibodies. Representative dot plot of one case is shown in (A) and (B) respectively
Ombretta Repetto https://orcid.org/0000-0002-0785-5066
Laura Gragnani https://orcid.org/0000-0001-6800-9149
Francesca Maria Rossi https://orcid.org/0000-0001-8438-1661
Riccardo Dolcetti https://orcid.org/0000-0003-1625-9853
Anna Linda Zignego https://orcid.org/0000-0002-8552-4166
Franco M. Buonaguro https://orcid.org/0000-0002-7491-7220
Agostino Steffan https://orcid.org/0000-0002-6320-3054
REFERENCES
1. Ryerson AB, Eheman CR, Altekruse SF, et al. Annual report to the nation on the status of cancer, 1975–2012, featuring the increasing incidence of liver cancer. Cancer. 2016;122:1312-1337.
2. De Re V, Caggiari L, Simula MP, De Vita S, Sansonno D, Dolcetti R. B-cell lymphomas associated with HCV infection. Gastroenterology. 2007;132:1205-1207.
3. Sansonno D, Carbone A, De Re V, Dammacco F. Hepatitis C virus infection, cryoglobulinaemia, and beyond. Rheumatology(Oxford). 2007;46:572-578.
4. Prokunina-Olsson L, Muchmore B, Tang W, et al. A variant upstream of IFNL3 (IL28B) creating a new interferon gene IFNL4 is associated with impaired clearance of hepatitis C virus. Nat Genet. 2013;45:164-171. 5. Xiao W, Jiang LF, Deng XZ, et al. PD-1/PD-L1 signal pathway
partic-ipates in HCV F protein-induced T cell dysfunction in chronic HCV infection. Immunol Res. 2016;64:412-423.
6. Kim PS, Ahmed R. Features of responding T cells in cancer and chronic infection. CurrOpinImmunol. 2010;22:223-230.
7. Blank C, Gajewski TF, Mackensen A. Interaction of PD-L1 on tumor cells with PD-1 on tumor-specific T cells as a mechanism of immune evasion: implications for tumor immunotherapy. Cancer
ImmunolImmunother. 2005;54:307-314.
8. Zamani MR, Aslani S, Salmaninejad A, Javan MR, Rezaei N. PD-1/ PD-L and autoimmunity: A growing relationship. Cell Immunol. 2016;310:27-41.
9. Li Z, Li NA, Li F, et al. Immune checkpoint proteins PD-1 and TIM-3 are both highly expressed in liver tissues and correlate with their gene polymorphisms in patients with HBV-related hepatocellular carcinoma. Med Baltim. 2016;95:e5749.
10. El-Khoueiry AB, Sangro B, Yau T, et al. Nivolumab in patients with advanced hepatocellular carcinoma (CheckMate 040): an open-la-bel, non-comparative, phase 1/2 dose escalation and expansion trial. Lancet. 2017;389:2492-2502.
11. Goodman A, Patel SP, Kurzrock R. PD-1-PD-L1 immune-checkpoint blockade in B-cell lymphomas. Nat Rev Clin Oncol. 2017;14:203-220. 12. Cox MA, Nechanitzky R, Mak TW. Check point inhibitors as ther-apies for infectious diseases. Curr Opin Immunol. 2017;48:61-67. 10.1016/j.coi.2017.07.016
13. Pervolaraki K, Rastgou Talemi S, Albrecht D, et al. Differential in-duction of interferon stimulated genes between type I and type III interferons is independent of interferon receptor abundance. PLoS
Pathog. 2018;14:e1007420.
14. Prokunina-Olsson L. Genetics of the human interferon lambda re-gion. J Interferon Cytokine Res. 2019;39:599-608.
15. Bamford CGG, Aranday-Cortes E, Filipe IC, et al. A polymorphic residue that attenuates the antiviral potential of interferon lambda 4 in hominid lineages. PLoS Pathog. 2018;14:e1007307.
16. Chinnaswamy S. Gene-disease association with human IFNL locus polymorphisms extends beyond hepatitis C virus infections. Genes
Immun. 2016;17:265-275.
17. De Re V, De Zorzi M, Caggiari L, et al. HCV-related liver and lymph-oproliferative diseases: association with polymorphisms of IL28B and TLR2. Oncotarget. 2016;7:37487-37497.
18. O’Brien TR, Pfeiffer RM, Paquin A, et al. Comparison of functional variants in IFNL4 and IFNL3 for association with HCV clearance. J
Hepatol. 2015;63:1103-1110.
19. Ansari MA, Aranday-Cortes E, Ip CL, et al. Interferon lambda 4 im-pacts the genetic diversity of hepatitis C virus. eLife. 2019;8. 20. Vergara C, Thio CL, Johnson E, et al. Multi-ancestry genome-wide
association study of spontaneous clearance of hepatitis C virus.
Gastroenterology. 2019;156(1496–1507):e7.
21. Tanaka Y, Nishida N, Sugiyama M, et al. Genome-wide association of IL28B with response to pegylated interferon-alpha and ribavirin therapy for chronic hepatitis C. NatGenet. 2009;41:1105-1109. 22. Prokunina L, Castillejo-López C, Öberg F, et al. A regulatory
poly-morphism in PDCD1 is associated with susceptibility to systemic lupus erythematosus in humans. Nat Genet. 2002;32:666-669. 23. Torre LA, Sauer AMG, Chen MS, Kagawa-Singer M, Jemal A, Siegel
RL. Cancer statistics for Asian Americans, Native Hawaiians, and Pacific Islanders, 2016: Converging incidence in males and females.
CA Cancer J Clin. 2016;66:182-202.
24. Petruzziello A, Loquercio G, Sabatino R, et al. Prevalence of Hepatitis C virus genotypes in nine selected European countries: A systematic review. J Clin Lab Anal. 2019;33.
25. European Association for the Study of the Liver. Electronic address: easloffice@easloffice.eu, European Association for the Study of the Liver. EASL Clinical Practice Guidelines: Management of hepatocel-lular carcinoma. J Hepatol 2018;69:182-236.
26. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms.
Blood. 2016;127:2375-2390.
27. Pezzuto F, Izzo F, Buonaguro L, et al. Tumor specific mutations in TERT promoter and CTNNB1 gene in hepatitis B and hepatitis C related hepatocellular carcinoma. Oncotarget. 2016;7(34):54253-54262. 28. Xiao W, Zhang Q, Deng XZ, et al. Genetic variations of IL-28B and
PD-1 are in association with the susceptibility and outcomes of HCV infection in Southeast China. Infection, Genetics and Evolution. 2015;32:89-96.
29. Qian BP, Wang XQ, Qiu Y, Jiang H, Ji ML, Jiang J. An exon poly-morphism of programmed cell death 1 gene is associated with both the susceptibility and thoracolumbar kyphosis severity of ankylosing spondylitis in a Chinese Han population. J OrthopSci. 2013;18:514-518.
30. Bertsias GK, Nakou M, Choulaki C, et al. Genetic, immunologic, and immunohistochemical analysis of the programmed death 1/ programmed death ligand 1 pathway in human systemic lupus ery-thematosus. Arthritis Rheum. 2009;60:207-218.
31. Kong EK, Prokunina-Olsson L, Wong WH, et al. A new haplotype of PDCD1 is associated with rheumatoid arthritis in Hong Kong Chinese. Arthritis Rheum. 2005;52:1058-1062.
32. Nielsen C, Hansen D, Husby S, Jacobsen BB, Lillevang ST. Association of a putative regulatory polymorphism in the PD-1 gene with susceptibility to type 1 diabetes. Tissue Antigens. 2003;62:492-497.
33. Da LS, Zhang Y, Zhang CJ, et al. The PD-1 rs36084323 A > G polymorphism decrease cancer risk in Asian: A meta-analysis.
PatholResPract. 2018;214:1758-1764.
34. Shamsdin SA, Karimi MH, Hosseini SV, et al. Associations of ICOS and PD.1 gene variants with colon cancer risk in the iranian popula-tion. Asian PacJ. Cancer Prev. 2018;19:693-698.
35. Tan D, Sheng L, Yi QH. Correlation of PD-1/PD-L1 polymorphisms and expressions with clinicopathologic features and prognosis of ovarian cancer. Cancer Biomark. 2018;21:287-297.
36. Salmaninejad A, Khoramshahi V, Azani A, et al. PD-1 and can-cer: molecular mechanisms and polymorphisms. Immunogenetics. 2018;70:73-86.
37. Tang W, Chen S, Chen Y, et al. Programmed death-1 polymorphisms is associated with risk of esophagogastric junction adenocarcinoma
